Dynamic O-(2-¹⁸F-fluoroethyl)-L-tyrosine positron emission tomography differentiates brain metastasis recurrence from radiation injury after radiotherapy

Abstract

Radiotherapy, particularly stereotactic radiosurgery (SRS), has become an increasingly important treatment option for the initial management of patients with brain metastases.¹ The efficacy of SRS, when used alone or combined with whole-brain radiation therapy (WBRT) has been demonstrated in phase 3 studies and shown a 12-month local control of 70%–90%.^2–4

Currently, conventional contrast-enhanced serial MR imaging is the method of choice for follow-up after treatment of metastatic brain tumors. In many patients, however, differentiating local recurrent brain metastasis from radiation injury (usually radionecrosis after radiotherapy, ie, SRS and/or WBRT) is difficult when using contrast-enhanced standard MRI.⁵ A radiation injury typically shows delayed occurrence and has been reported to occur in up to 25% of patients after the completion of radiotherapy.⁶ Depending on the irradiated volume receiving a specific radiation dose, the risk of radionecrosis may be as high as 50%.⁶ Furthermore, clinical monitoring may also be challenging since both recurrent brain metastasis and radiation injury are characterized by similar neurological symptoms and signs.⁷

In order to overcome this highly relevant clinical problem, metabolic PET imaging using 2-[¹⁸F]fluoro-2-deoxy-D-glucose (¹⁸F-FDG) PET has been tested, but the high physiological glucose consumption of the brain and the variable glucose uptake of metastatic brain lesions limit its use.^8,9 Recently, more encouraging results were obtained with the use of amino acid PET tracers including ¹¹C-methyl-L-methionine (¹¹C-MET), O-(2-[¹⁸F]fluoroethyl)-L-tyrosine (¹⁸F-FET) and L-3,4-dihydroxy-6-¹⁸F-fluoro-phenylalanine (¹⁸F-FDOPA).^10–15 In these studies, data evaluation was based predominantly on static PET parameters (eg, tumor-to-brain ratios).

In glioma patients, the use of dynamic ¹⁸F-FET PET scans has gained great interest since it has been demonstrated that the evaluation of kinetic parameters (eg, patterns of time-activity curves, time-to-peak values) add valuable information in terms of differentiation between low-and high-grade glioma,^16–22 prognostication in untreated glioma patients,^22–24 and the distinction of posttreatment changes from glioma recurrence (eg, pseudoprogression).^25,26

In a pilot study of our group, we recently demonstrated that the combination of static and dynamic ¹⁸F-FET PET parameters in patients with brain metastasis is superior to static PET parameters alone for differentiating between brain metastasis recurrence and radiation injury.¹² However, due to the pilot character of that study, a relatively low number of patients with a short follow-up were examined. In the present study, patients in whom no histological confirmation of diagnosis could be obtained were re-evaluated with respect to a longer follow-up period (median follow-up, 16 mon); moreover, we included additional patients in order to increase the patient population. Furthermore, we provided a more objective evaluation of kinetic data (see Patients and Methods section) in the present study.

Materials and Methods

Patients

Sixty-two patients with metastatic brain tumors (mean age, 55 ± 11 y; range, 17–79 y; 48 women and 14 men), each having at least one contrast-enhancing lesion (n = 76) on cerebral MRI, were included in this retrospective study (Table 1). The patients were consecutively referred (from 2005 to 2014) to our institute to differentiate local recurrent brain metastasis from radiation injury using ¹⁸F-FET PET because of suspicious MRI findings such as new contrast-enhancing lesions or the progression of contrast enhancement at the site of the initial metastasis All patients had been previously treated with radiotherapy (RT) (median time between RT and suspicious MRI findings, 14 mo; range, 3–64 mo), predominantly with SRS alone (32 patients (51%); range of surface radiation dose, 16–25 Gy/50–80% isodose level), or with WBRT (25–30 Gy, 1.8–2.0 Gy per fraction) in combination with SRS (21 patients (34%); range of surface radiation dose, 16–25 Gy/50–80% isodose level). The median time interval between RT and PET scanning was 13.5 months (range, 3–64 mo). All patients gave written informed consent before each ¹⁸F-FET PET investigation. The local ethics committee approved the evaluation of retrospectively collected patient data.

Histopathogical results for definite diagnosis were available in 26 lesions (34%) from 25 patients. In the remaining patients (ie, 37 patients with 50 lesions [66%]), diagnosis of recurrent brain metastasis or radiation injury was based on their follow-up (ie, clinical course and results of follow-up MRI). Recurrent disease was anticipated if a new contrast-enhancing lesion appeared at exactly the same site as the treated metastasis after initial complete response or the treated metastasis showed progression in size during follow-up according to Response Assessment in Neuro-Oncology (RANO) criteria for brain metastasis²⁷ (increase of >20% in the pretreated volume on contrast enhanced T1-weighted MR images) and new neurological deficits or exacerbation of existing neurological symptoms prompting a change in treatment. Radiation injuries in the tissue were assumed when (i) the lesions showed spontaneous shrinkage or remained stable in size on contrast-enhanced MRI during follow-up (median follow-up, 16 mo; range, 3–63 mo); (ii) neurological deficits remained unchanged; (iii) and no new neurological symptoms occurred.

PET Imaging With ¹⁸F-FET

The amino acid ¹⁸F-FET was produced as described previously.^28,29 According to the German guidelines for brain tumor imaging using labelled amino acid analogues, all patients fasted for at least 12 hours before PET scanning.³⁰ Dynamic PET studies were acquired up to 50 minutes after intravenous injection of approximately 200 MBq ¹⁸F-FET on an ECAT EXACT HR+ scanner (Siemens Medical Systems, Inc.) in 3-dimensional mode (32 rings; axial field of view, 15.5 cm). The emission recording consisted of 16 time frames (time frames 1–5: 1 min, 6–10: 3 min, and 11–16: 5 min) covering the period up to 50 minutes postinjection. For attenuation correction, transmission was measured with 3 ⁶⁸Ge/⁶⁸Ga rotating line sources. After correction for random and scattered coincidences as well as dead time, 63 image planes were iteratively reconstructed (OSEM, 6 iterations, 16 subsets) using the ECAT 7.2 software. The reconstructed image resolution was approximately 5.5 mm.

PET Data Analysis

¹⁸F-FET uptake in the tissue was expressed as standarized uptake value (SUV) by dividing the radioactivity (kBq/mL) in the tissue by the radioactivity injected per gram of body weight. PET and MR images were co-registered using dedicated software (MPI tool version 6.48; ATV). The fusion results were inspected and, if necessary, adapted based on anatomical landmarks. The regions-of-interest (ROI) analysis and the calculation of PET tumor volumes were based on the summed PET data from 20–40 minutes postinjection. The transaxial slices showing the highest tracer accumulation in the tumors were chosen for ROI analyses. The uptake in the unaffected brain tissue was determined by a larger ROI placed on the contralateral hemisphere in an area of normal-appearing brain tissue including white and gray matter.³⁰ Mean and maximum amino acid uptake in the tumor was determined by a 2-dimensional autocontouring process using a tumor-to-brain ratio (TBR) of 1.6 as described previously.³¹ Maximum and mean tumor-brain-ratios (TBR_max, TBR_mean) were calculated by dividing the mean SUV of these tumor ROIs by the mean SUV of normal brain in the PET scan. The calculation of PET tumor volumes was determined by a 3D autocontouring process using a threshold of 1.6 with PMOD (Version 3.505, PMOD Technologies Ltd.).

Furthermore, time-activity curves (TACs) of ¹⁸F-FET uptake in the tumor were generated by the application of a spherical volume-of-interest (VOI) with a volume of 2 mL centered on maximal tumor uptake to the entire dynamic dataset. This VOI approach differed from the evaluation in our pilot study in order to provide a simplified TAC evaluation focused on an area with maximum radioactivity in the tumor.^25,26 TAC of the brain tissue was generated by a reference ROI in the unaffected brain tissue (as described above). Time-to-peak (TTP; time in minutes from the beginning of the dynamic acquisition up to the maximum SUV of the lesion) was determined. In extension to our previous study,¹² we quantified the slope of the TAC in the late phase of ¹⁸F-FET uptake by fitting a linear regression line to the late phase of the curve (20–50 min postinjection). The slope was expressed in change of SUV per hour.

Statistical Analysis

Descriptive statistics are provided as mean and standard deviation and/or median and range. To compare 2 different groups, the Student t test for independent samples was used. The Mann-Whitney rank-sum test was used when variables were not distributed normally.

The diagnostic performance of ¹⁸F-FET uptake, as determined by TBR_max and TBR_mean as well as TTP to identify brain metastasis recurrence after RT, was assessed by receiver-operating-characteristic (ROC) curve analyses using histological confirmation or clinical course as reference. Decision cutoff was considered optimal when the product of paired values for sensitivity and specificity reached its maximum. In addition, the area under the ROC curve (AUC), its standard error, and level of significance were determined as a measure of diagnostic quality. The diagnostic performance of ¹⁸F-FET TAC slopes alone and in combination with the corresponding TBRs was evaluated by chi-square or Fisher exact test for 2 × 2 contingency tables. Using an approximative procedure, different slope cutoffs were tested in combination with the TBRs to determine the best diagnostic performance.

P values ≤.05 were considered significant. Statistical analyses were performed using SigmaPlot software (SigmaPlot Version 11.0, Systat Software Inc.) and PASW Statistics software (Release 22.0.0, SPSS Inc.).

Results

Histological confirmation of diagnosis was obtained in 25 patients with 26 lesions, and histopathology showed viable tumor tissue in 21 lesions (Fig. 1) and radiation-induced changes (ie, necrosis) with no/single viable tumor cells in 5 lesions (Fig. 2). Thirty-five lesions in 27 patients without histological evaluation were classified as radiation injury because the patients exhibited stable neurological symptoms and no significant enlargement of the lesion on follow-up MR images after a median of 16 months (range, 3–63 min). Predominantly due to clinical deterioration of neurological symptoms (ie, reduction of the Karnofsky performance index <60% and/or subsequent death) as well as progression in size on contrast-enhanced MRI during follow-up (median, 4 mo; range, 0–75 mo), 15 lesions in 10 patients were classified as recurrent brain metastases (Supplementay material, Table S1). Regarding radiotherapy schemes, we found a weak significance (P = .04) that in the group of patients which were treated with radiosurgery alone a radiation injury occurs more often than in the group treated with both WBRT and radiosurgery. Within the subgroups of the originating primary tumor (ie, lung cancer, breast cancer, miscellaneous), the radiation injury rate was not statistically increased. We found a trend of an increased radiation injury rate in older patients treated with radiosurgery only (P = .066). In the group of patients receiving radiosurgery alone, a smaller FET PET tumor volume was associated with a higher radiation injury rate (P = .019).

Fig. 1.

58-year old patient (#4) with a brain metastasis of a non–small cell lung cancer. Twenty-nine months after stereotactic radiosurgery (SRS), MRI suggests tumor recurrence. In line with MRI findings, ¹⁸F-FET PET shows increased metabolic activity (TBR_mean = 2.2), and the slope of the late phase of the TAC (20–50 min postinjection) is −0.72 SUV/h. Histology after resection was consistent with a recurrent brain metastasis (hematoxylin and eosin stain; original magnification ×400; scale bar, 1000 µm).

Open in new tabDownload slide

Fig. 2.

61-year old patient (#52) with a brain metastasis of a non–small cell lung cancer. Nine months after stereotactic radiosurgery (SRS), MRI suggests tumor recurrence. In contrast, ¹⁸F-FET PET shows no increased metabolic activity (TBR_mean = 1.6), and the slope of the TAC is 0.38 SUV/h. Histology after stereotactic biopsy revealed fibrohyaline-thickened blood vessel walls and amorphic tissue necroses with rims of macrophages. The findings are compatible with actinic angiopathy and radiation necrosis (hematoxylin and eosin stain; original magnification ×100; scale bar, 1000 µm).

Open in new tabDownload slide

Comparison of Uptake Indices for Recurrent Brain Metastasis and Radiation Injuries

Both TBR_max and TBR_mean were significantly higher in recurrent metastases (n = 36) than in radiation injuries (n = 40) (TBR_max 3.3 ± 1.0 vs 2.2 ± 0.4, P < .001; TBR_mean 2.2 ± 0.4 vs 1.7 ± 0.3, P < .001).

Receiver Operating Characteristic Analysis of Tumor-brain Ratio Values

The diagnostic accuracy of TBR values for the correct identification of recurrent brain metastases reached 84% using TBR_max (AUC, 0.883 ± 0.04; sensitivity, 83%; specificity, 85%; cutoff, 2.55; P = .001) and 87% using TBR_mean (AUC, 0.919 ± 0.03; sensitivity, 86%; specificity, 88%; cutoff, 1.95; P = .001) (Table 2).

Diagnostic Performance of Kinetic ¹⁸F-FET PET Parameters

ROC analysis of the slope in late-phase ¹⁸F-FET uptake revealed a diagnostic accuracy of 64% (AUC, 0.676 ± 0.06; sensitivity, 68%; specificity, 61%; cutoff, 0.125 SUV/h; P = .008). For TTP, ROC analysis revealed a diagnostic accuracy of 66% (AUC, 0.758 ± 0.06; sensitivity, 58%; specificity, 73%; cutoff, 32.5 min; P = .001) (Table 2).

Combined Analysis of Tumor-brain Ratio and Kinetic ¹⁸F-FET PET Parameters

The highest diagnostic accuracy (88%) for diagnosing recurrent brain metastases was obtained when both conditions TBR_mean > 1.95 and presence of a slope <0.37 SUV/hour in the late phase of ¹⁸F-FET uptake were met (sensitivity, 83%; specificity, 93%; P < .001) (Table 2). For the subset of patients with histological confirmation of diagnosis (recurrent disease or radiation injury), the combined analysis on diagnostic performance using TBR_mean and the kinetic ¹⁸F-FET PET parameter slope revealed the same diagnostic accuracy (88%). In contrast, presence of a TBR_mean > 1.95 in combination with a TTP <32.5 minutes revealed a diagnostic accuracy of 78% (sensitivity, 58%; specificity, 95%; P < .001).

In order to rule out a bias by mixing dependent and independent samples (patients with single or multiple lesions), a by-patient–stratified subanalysis using the largest lesion only in each patient revealed a similar diagnostic accuracy of 90% (sensitivity, 90%; specificity, 91%; positive predictive value, 90%; negative predictive value, 91%; P < .001).

Discussion

Radiation injury is a well-recognized complication of radiotherapy, and its incidence is growing due to the increased number of SRS procedures worldwide and the improvement in patient survival. Today, the use of SRS is a widely accepted treatment option for cerebral metastases,¹ either as a single modality or in combination with WBRT. Radiation injury (eg, radiation necrosis) on follow-up standard MRI is characterized by alterations in T2-weighted images and changing patterns of contrast enhancement and is often indistinguishable from local tumor recurrence.^32–34 The current gold standard for distinguishing tumor recurrence from radiation injury remains biopsy, which has an accuracy rate >95%.^35,36 Biopsy, however, is invasive, can yield false-negative results, and has potential complications such as infection, procedure-associated new neurological problems, and hematoma—although the risk of permanent complications is relatively low.^35,36 Therefore, a noninvasive imaging technique that provides valuable additional information is desirable—for instance, the use of PET.

The classic and most common PET tracer for oncologic imaging, both neurologic and non-neurologic, used to be ¹⁸F-FDG. ¹⁸F-FDG is accumulated in the majority of tumors due to increased energy demand and the consequent elevated glucose metabolism. ¹⁸F-FDG uptake has been well characterized for tumors outside the brain and has also been used for brain-tumor imaging for many years. The use of ¹⁸F-FDG in brain-tumor imaging, however, is limited by high background uptake in brain tissue (especially in brain metastases due to variable glucose consumption),^8,9 although potentially higher diagnostic accuracy may be obtained by dual-phase imaging.³⁷ Of note, a limitation of that approach for routine application is the long time interval between PET scans (range, 2–5.7 h).

In the present study, therefore, we evaluated the diagnostic value of dynamic ¹⁸F-FET PET for differentiaton of recurrent brain metastasis from radiation injury. Findings of the present study were based on a larger sample size (inclusion of 31 additional patients) and longer follow-up times than our previous preliminary report.¹²

Patients in our sample had been predominantly treated by radiosurgery, and standard follow-up MRI suggested recurrent brain metastasis. The results of our study demonstrated that even static ¹⁸F-FET PET parameters alone could distinguish between recurrent brain metastasis and radiation injury with a high diagnostic accuracy (87% when using TBR_mean). In contrast to our previous study,¹² the addition of the kinetic analysis of ¹⁸F-FET tracer uptake to static parameters showed only a small improvement of overall diagnostic accuracy, but both specificity and the positive predictive value showed marked improvement. To the best of our knowledge, this amino acid PET study includes the highest number of patients in whom the differentiation between recurrent brain metastasis and radiation-induced changes was evaluated.

Currently, only a few studies have evaluated the diagnostic performance of amino acid tracers in the setting of differentiating recurrent disease from radiation injury in patients with brain metastases treated with radiotherapy. For instance, ¹¹C-MET PET may be effective for differentiating recurrent metastatic brain tumor from radiation-induced changes as the simple calculation of TBRs demonstrated sensitivity and specificity > 70%.^{10,11 18}F-FDOPA PET has been shown to differentiate recurrent or progressive brain metastasis from radiation-induced changes with high sensitivity (81%) and specificity (84%).¹³ Another study has reported an accuracy of ¹⁸F-FDOPA PET of 91% for differentiating radiation-induced changes from progressive disease in patients with brain metastases after stereotactic radiosurgery, which was superior to the accuracy derived from perfusion-weighted MRI (76%).¹⁴ A similar diagnostic performance has also been reported in our pilot ¹⁸F-FET PET study. Using TBRs in combination with the evaluation of ¹⁸F-FET kinetic studies, ¹⁸F-FET PET differentiated local recurrent brain metastasis from radiation-induced changes with an accuracy of 93%.¹² In the present study, we replaced the visual curve pattern analysis used in our previous study with a more objective evaluation of kinetic data (ie, slope of the TAC in the late phase of ¹⁸F-FET uptake 20–50 min postinjection). Using this combinational approach in the present study, we found only slightly increased diagnostic accuracy compared with the static approach (88% vs 87%). On the other hand, we observed marked improvement of specificity (93% vs 88%) and positive predictive value (91% vs 86%) (Table 2).

Our data suggest that the occurrence of radiation injury was more frequent in patients receiving radiosurgery alone than in those treated with radiosurgery and WBRT. Furthermore, in the group of patients receiving radiosurgery alone, a smaller ¹⁸F-FET PET tumor volume was associated with a higher radiation injury rate (P = .019). However, these findings  contradict other reported findings. It has been demonstrated that both larger tumor size and prior WBRT are associated with a higher rate of radionecrosis.³⁸ On the other hand, it has to be noted that the statistical difference for the higher rate of radiation injury in the group treated with radiosurgery alone is weak (P = .04), and the number of patients in the underlying dataset is too low to derive meaningful conclusions.

One could argue that the present study suffers from several limitations. First, because multiple lesions occurred frequently and biopsy was not available for all lesions, radiological/clinical criteria had to be used for the final diagnosis in two-thirds of the lesions. Second, the relatively long scanning time (50 min) may cause artifacts due to the patient's head motion. This, however, was not the case in our study. Third, dynamic scanning is more laborious and time-consuming. It is still unclear whether the improvement of diagnostic performance of dynamic scanning justifies the additional costs. This issue needs to be further evaluated in a prospective trial with a larger series of patients.

In conclusion, the initial results of our pilot study¹² could be successfully reconfirmed. Our findings show that ¹⁸F-FET PET is a useful method in a clinically challenging situation, especially when conventional MRI is inconclusive. The detection of recurrent brain metastasis with high accuracy is essential for optimizing patient counseling as well as the treatment strategy for each individual patient. Our data suggest that this approach achieves an accuracy that is sufficient to influence clinical decision-making and may therefore help to reduce the number of invasive diagnostic interventions and overtreatment for a considerable number of seriously ill patients with brain metastases. A larger prospective study is warranted to confirm the clinical usefulness of ¹⁸F-FET PET-derived imaging parameters for differentiating local recurrent brain metastasis from radiation injury after radiation therapy.

Supplementary Material

Supplementary material is available at Neuro-Oncology Journal online (http://neuro-oncology.oxfordjournals.org/).

Funding

None.

Acknowledgments

The authors thank Suzanne Schaden, Elisabeth Theelen, Silke Frensch, Kornelia Frey, and Lutz Tellmann for assistance in the patient studies and Johannes Ermert, Silke Grafmüller, Erika Wabbals, and Sascha Rehbein for radiosynthesis of ¹⁸F-FET.

Conflict of interest statement. The authors disclosed no potential conflicts of interest.

References

Author notes